Consolidate parsing into common_functions

[gerritdm]

I ran into some of these parsing issues again when dealing with the
plots so I went ahead and tried to address some of the issues raised
in #111 (e.g. we're still ignoring unknown arguments in the sub-scripts
called in run_end_to_end.py).

This PR re-factors the adding of parsed arguments into a set
of functions that add related arguments. Each script can then simply
use some of these function as parent parsers when setting up the
ArgumentParser for that script.

Note: there are still some issues with how arguments are passed through
from run_end_to_end to the other scripts. For instance, the scenario
name is required for run_scenario.py but in theory not for
run_end_to_end.py. However, run_end_to_end.py calls run_scenario.py, so
you will get an error message if you don't provide it in
run_end_to_end.py and the error message will be a somewhat confusing
combination of run_end_to_end.py and run_scenario.py

One solution could be to get rid of scenario_id and make scenario a
required argument. The value can be either the ID or the name, and we
can add some functionality to smartly determine what is provided and
get the equivalent name/id from the db (if needed).

[gerritdm]

2 things that we might want to address in this PR as well:

• remove reverting to a hard-coded default scenario_location in gridpath.common_functions.determine_scenario_directory() and simply use the default value from the argument parser. We might want to get rid of the function entirely and simply call os.path.join(scenario_location, scenario) whenever needed.
• similarly for the database, I think it might be more clear to add the default db_path to the argument parser definition, rather than assigning it in db.common_functions.connect_to_database.

[anamileva]

Looks good.